Welcome Guest!
Create an Account
login email:
password:
site searchwhere to watchcontact usabout usadvertise with ushelp
Message Board
BobcatAttack.com Message Board
Ohio Football Recruiting
Topic:  Using recruiting rankings to evaluate comparative talent levels

Topic:  Using recruiting rankings to evaluate comparative talent levels
Author
Message
L.C.
General User

Member Since: 8/31/2005
Location: United States
Post Count: 10,470

Status: Offline
  Message Not Read  Using recruiting rankings to evaluate comparative talent levels
   Posted: 11/8/2023 8:06:05 AM 
I have long maintained that, while there was never a doubt that recruiting rankings were accurate at the 4-5 star level, recruiting rankings at the G5 level were random and meaningless, i.e. "for amusement only". This was because the recruiting services put virtually no effort into evaluating those athletes. In recent years, ESPN and Rivals have acknowledged that they ignore G5, and have stopped rating G5 athletes, except for those that they accidentally evaluated thinking they might be destined for P5 schools.

247Sports also acknowledged that their ratings of G5 athletes were not good, but instead of stopping, they vowed to get better at it. I have noticed that the rankings by 247Sports of G5 athletes have steadily improved. As recently at 2017-2018 they were still not great, but the last few years seem to have been much, much better. Thus, I decided to use their rankings to predict which teams would be the best, and compared the winning records of the teams which were predicted to have been in the top half to the records of the teams predicted be in the bottom half. For years up to 2019, there was no predictive value at all, however, the last few years, teams predicted to be in the top half have significantly outperformed teams predicted to be in the bottom half.

For this year, the predicted levels of talent (based on a composite of recruiting the last 5 years), and the current TeamRankings projected conference wins are:
1. Toledo 83.05 (7.5-0.5)
2. WMU 82.07 (3.8-4.2)
3. Miami 81.31 (6.6-1.4)
4. Ohio 81.06 (5.7-2.3)
5. NIU 80.97 (4.5-3.5)
6. CMU 80.69 (3.4-4.6)
7. Kent 80.67 (0.5-7.5)
8. Buffalo 80.54 (3.8-4.2)
9. BG 80.46 (4.6-3.4)
10. Ball State 80.28 (3-5)
11. EMU 80.22 (3.2-4.8)
12. Akron 80.03 (1.4-6.6)

Conclusions:
1. It's not a perfect match, but it would seem that 247Sports has definitely improved their recruiting rankings. The recruiting rankings do seem to have some predictive value now.
2. Note that the MAC is very close, and there is not a great deal of difference between #3 and #9. That's why the unexpected happens so often in MAC games.
3. There are lots of reasons why actual results never completely match recruiting rankings. For example, coaching affects outcomes. Keeping the best players happy and out of the portal affects things. Adding good players from the portal affects things.

Predictions for future years:
Things will change slightly, based on how recruiting finishes up this year, but Ohio's recruiting classes have been especially strong this year and last year, so Ohio's predicted outcomes will be high all the way to 2028, unless the next few years are awful. Toledo will still out front for next year, but other teams will be closer, and the entire MAC will be closer as the bottom teams are all higher.

“We have two ears and one mouth so that we can listen twice as much as we speak.” ― Epictetus

Back to Top
  
Jeff McKinney
Moderator

Member Since: 11/12/2004
Post Count: 6,121

Status: Offline
  Message Not Read  RE: Using recruiting rankings to evaluate comparative talent levels
   Posted: 11/8/2023 11:51:06 AM 
Excellent analysis. Thanks for taking the time.
Back to Top
  
CoachPMac
General User

Member Since: 4/11/2023
Post Count: 32

Status: Offline
  Message Not Read  RE: Using recruiting rankings to evaluate comparative talent levels
   Posted: 11/8/2023 12:21:40 PM 
L.C. wrote:
I have long maintained that, while there was never a doubt that recruiting rankings were accurate at the 4-5 star level, recruiting rankings at the G5 level were random and meaningless, i.e. "for amusement only". This was because the recruiting services put virtually no effort into evaluating those athletes. In recent years, ESPN and Rivals have acknowledged that they ignore G5, and have stopped rating G5 athletes, except for those that they accidentally evaluated thinking they might be destined for P5 schools.

247Sports also acknowledged that their ratings of G5 athletes were not good, but instead of stopping, they vowed to get better at it. I have noticed that the rankings by 247Sports of G5 athletes have steadily improved. As recently at 2017-2018 they were still not great, but the last few years seem to have been much, much better. Thus, I decided to use their rankings to predict which teams would be the best, and compared the winning records of the teams which were predicted to have been in the top half to the records of the teams predicted be in the bottom half. For years up to 2019, there was no predictive value at all, however, the last few years, teams predicted to be in the top half have significantly outperformed teams predicted to be in the bottom half.

For this year, the predicted levels of talent (based on a composite of recruiting the last 5 years), and the current TeamRankings projected conference wins are:
1. Toledo 83.05 (7.5-0.5)
2. WMU 82.07 (3.8-4.2)
3. Miami 81.31 (6.6-1.4)
4. Ohio 81.06 (5.7-2.3)
5. NIU 80.97 (4.5-3.5)
6. CMU 80.69 (3.4-4.6)
7. Kent 80.67 (0.5-7.5)
8. Buffalo 80.54 (3.8-4.2)
9. BG 80.46 (4.6-3.4)
10. Ball State 80.28 (3-5)
11. EMU 80.22 (3.2-4.8)
12. Akron 80.03 (1.4-6.6)

Conclusions:
1. It's not a perfect match, but it would seem that 247Sports has definitely improved their recruiting rankings. The recruiting rankings do seem to have some predictive value now.
2. Note that the MAC is very close, and there is not a great deal of difference between #3 and #9. That's why the unexpected happens so often in MAC games.
3. There are lots of reasons why actual results never completely match recruiting rankings. For example, coaching affects outcomes. Keeping the best players happy and out of the portal affects things. Adding good players from the portal affects things.

Predictions for future years:
Things will change slightly, based on how recruiting finishes up this year, but Ohio's recruiting classes have been especially strong this year and last year, so Ohio's predicted outcomes will be high all the way to 2028, unless the next few years are awful. Toledo will still out front for next year, but other teams will be closer, and the entire MAC will be closer as the bottom teams are all higher.


That’s not quite accurate about Rivals. In fact, if you take a look at the 2023 MAC ratings for HS/JUCO and the newly added Comprehensive Rankings, you’ll find Rivals actually had more players rated than any other site. You’ll also find the 2024 MAC commitments are more up to date at Rivals than any other site.

The pitfalls are currently in a handful of areas where the evaluator does not give as much attention to prospects deemed under 3 stars. However, the coverage for Midwest recruits is excellent, as a former coach handles those ratings.

An issue has cropped up after the late signing period. All committed prospects will be given a rating at the end of the first signing period, but don’t always receive one by the second signing period or later. What we’re seeing now is a higher number of JUCOs signing later and are not receiving a rating. On a side note, better JUCO prospects are available than ever before due most teams putting a heavier emphasis on the portal.


Last Edited: 11/8/2023 5:05:11 PM by CoachPMac
Back to Top
  
L.C.
General User

Member Since: 8/31/2005
Location: United States
Post Count: 10,470

Status: Offline
  Message Not Read  RE: Using recruiting rankings to evaluate comparative talent levels
   Posted: 11/8/2023 6:58:50 PM 
CoachPMac wrote:
That’s not quite accurate about Rivals.
...

To be honest, I stopped looking at Rivals for the most part once they stopped rating MAC recruits, except for MAC schools that have a Rivals forum, some years ago. I occasionally look up Ohio recruits there, and sometimes they are rated, but often not. Even then, the "Ohio" for team is not a link, so there is no way that I am aware of to find a list of those they show as committed to Ohio.

As far as the accuracy of Rivals ratings, I have only experience with ratings from 5-10 years ago. At that time, I found no correlation between ratings and performance on the field on any of the ratings services. It was a running joke I had that the lowest rated player in each class would be the one who turned out to be an impact player. Examples include 2012 Toran Davis and Blair Brown, 2013 Cleon Aloese and Durrell Wood, 2014 Maleek Irons, Papi White, Kylan Nelson, and Mitch Bonnstetter, 2016 Dylan Conner, 2017 Kai Caesar, Hagan Meservy.

I searched around Rivals, and did find this page for 2024 MAC rankings:
https://n.rivals.com/team_rankings/2024/midam/football/re...

For only four teams (Toledo, Kent, WMU, and CMU) can you find the individual recruits, and each recruit's ranking. It says that Ohio's recruits have an average of 1.88 stars, and that there are 4 3-star players. That means that the other 12 players have an average of 1.5 stars, which I presume means that 8 are 2 stars, and 4 are entirely unrated. Given the limited data available, I have no way to attempt any correlations to see if the accuracy has improved. Perhaps they have improved, but I have no way to know.

Jeff McKinney wrote:
Excellent analysis. Thanks for taking the time.

Considering how many times over the years I have posted data showing no correlation whatsoever between recruiting rankings and performance (at the MAC level), I was surprised to find that it seems to have changed. So far as I recall, this is the first time I've shown a correlation. We'll see if that correlation holds into the future, or if this was just a fluke.

Last Edited: 11/8/2023 9:02:55 PM by L.C.

“We have two ears and one mouth so that we can listen twice as much as we speak.” ― Epictetus

Back to Top
  
CoachPMac
General User

Member Since: 4/11/2023
Post Count: 32

Status: Offline
  Message Not Read  RE: Using recruiting rankings to evaluate comparative talent levels
   Posted: 11/10/2023 5:48:36 AM 
L.C. wrote:
CoachPMac wrote:
That’s not quite accurate about Rivals.
...

To be honest, I stopped looking at Rivals for the most part once they stopped rating MAC recruits, except for MAC schools that have a Rivals forum, some years ago. I occasionally look up Ohio recruits there, and sometimes they are rated, but often not. Even then, the "Ohio" for team is not a link, so there is no way that I am aware of to find a list of those they show as committed to Ohio.

As far as the accuracy of Rivals ratings, I have only experience with ratings from 5-10 years ago. At that time, I found no correlation between ratings and performance on the field on any of the ratings services. It was a running joke I had that the lowest rated player in each class would be the one who turned out to be an impact player. Examples include 2012 Toran Davis and Blair Brown, 2013 Cleon Aloese and Durrell Wood, 2014 Maleek Irons, Papi White, Kylan Nelson, and Mitch Bonnstetter, 2016 Dylan Conner, 2017 Kai Caesar, Hagan Meservy.

I searched around Rivals, and did find this page for 2024 MAC rankings:
https://n.rivals.com/team_rankings/2024/midam/football/re...

For only four teams (Toledo, Kent, WMU, and CMU) can you find the individual recruits, and each recruit's ranking. It says that Ohio's recruits have an average of 1.88 stars, and that there are 4 3-star players. That means that the other 12 players have an average of 1.5 stars, which I presume means that 8 are 2 stars, and 4 are entirely unrated. Given the limited data available, I have no way to attempt any correlations to see if the accuracy has improved. Perhaps they have improved, but I have no way to know.

Jeff McKinney wrote:
Excellent analysis. Thanks for taking the time.

Considering how many times over the years I have posted data showing no correlation whatsoever between recruiting rankings and performance (at the MAC level), I was surprised to find that it seems to have changed. So far as I recall, this is the first time I've shown a correlation. We'll see if that correlation holds into the future, or if this was just a fluke.


247 and On3 certainly have the ease of use features down. Rivals is still in the process of updating a system from the early 2000s. You can still view individual recruits with the recruit search feature, but it’s not as user friendly. Rivals will not create any more individual G5 team sites because they do not produce much money and in the end it’s all about the business.

The majority of evaluators from 5-10 years that made it seem like a “running joke” are now at either 247 or On3 btw.

I’ve added the 2023 and 2024 team rankings on the CSNbbs MAC main board for easier viewing.

Use whichever service suits you. I just wanted to touch on the Rivals part.


Last Edited: 11/10/2023 8:50:08 AM by CoachPMac
Back to Top
  
L.C.
General User

Member Since: 8/31/2005
Location: United States
Post Count: 10,470

Status: Offline
  Message Not Read  RE: Using recruiting rankings to evaluate comparative talent levels
   Posted: 11/10/2023 9:31:43 AM 
CoachPMac wrote:
247 and On3 certainly have the ease of use features down. Rivals is still in the process of updating a system from the early 2000s. You can still view individual recruits with the recruit search feature, but it’s not as user friendly. Rivals will not create any more individual G5 team sites because they do not produce much money and in the end it’s all about the business.

The majority of evaluators from 5-10 years that made it seem like a “running joke” are now at either 247 or On3 btw.

I’ve added the 2023 and 2024 team rankings on the CSNbbs MAC main board for an easier viewing fashion.

Use whatever site suits you. I just wanted to touch on the Rivals part.

I did know that you could look up recruits individually at Rivals, though not at ESPN. I did not know that Rivals had restarted having MAC recruiting rankings, apparently back in 2021. I have no idea if Rivals or ESPN is now more accurate than they used to be at the G5 level. Both are, without question, accurate at the P5 level, and I'm not questioning that.

The main point of my post was specifically related to 247Sports. Since I have always considered them (along with Rivals, Scout.com, and ESPN) to be mostly random, I was surprised when I noticed that, not counting kickers, four True Freshmen are playing for Ohio, and all four are among the top 5 recruits in last year's class. In the past it was more normal for most of the True Freshmen who play to be at the bottom of the class. I therefore decided to do a test to see if overall MAC recruiting rankings at 247 were correlated to actual team wins, and to my surprise, they are.

Let's go back to that starting point, and see how Rivals and ESPN rated the four True Freshmen who are playing:
Chase Hendricks Rivals=3 stars/5.5, ESPN=3 stars/74, 247=86
Kobi Gorman Rivals=3 stars/5.5, ESPN=3 stars/76, 247=86
Ricky Hunt, Jr. Rivals=3 stars/5.5, ESPN=3 stars/72 247=86
Mason Williams Rivals=2 stars,5.4, ESPN=NR, 247=84

...and how they rated the other Ohio recruits from 2023:
Andrew Marshall Rivals=2 stars/5.4, ESPN=NR, 247=86
Danny Novickas Rivals=2 stars/5.4, ESPN=NR, 247=83
Jayvian Crable Rivals=2 stars/5.3, ESPN=NR, 247=82
Byron Pearson Rivals=2 stars/5.3, ESPN=NR, 247=82
Kendall Bannister Rivals=2 stars/5.2, ESPN=NR, 247=82
Jacob Winters Rivals=2 stars/5.4, ESPN=3 stars/72, 247=81
Desmond Duffy Rivals=2 stars/5.4, ESPN=3 stars/77, 247=80
Donovan Walker Rivals=2 stars/5.2, ESPN=NR, 247=80
Aidan Johnson Rivals=2 stars/5.3, ESPN=NR, 247=80
Owen DiFranco Rivals=2 stars/5.3, ESPN=NR, 247=80
Jalen Thompson Rivals=2 stars/5.3, ESPN=NR, 247=77
Brady Sestili Rivals=2 stars/5.2, ESPN=NR, 247=77
Gianni Spetic Rivals=2 stars/5.2, ESPN=NR, 247=75

Rivals looks very good. Their top three recruits (all 5.5) are all included in the 4 True Freshmen who are playing. The other true Freshmen who is playing is one of the 5 players that they rated next best, at 5.4. ESPN doesn't fare quite as well on this test. Their top-rated player, Desmond Duffy, is not playing, and one of the True Freshmen who is playing is NR. Still, of their five top rated players, 3 are playing, so there is still some correlation.

I did use ratings from the last five years to test the correlation between 247's rankings to team wins, and found a strong correlation. I would do the same for Rivals, but unfortunately I can not. MAC team rankings are only available back to 2021, and you need to go back further, because many/most starters are from the classed of 2019 or 2020, with even some from 2018 or before, due to Covid.

While 247Sports and Rivals do not completely agree, there are sometimes distinct differences, in this case, primarily related to the last two years. Here are team recruiting standings for Ohio in the MAC for the last four years. 2021 and 2022 they agree, but 2023 is quite different:
2021 Class: Rivals=>Ohio is #9 in the MAC; 247Sports=>Ohio is #9 in the MAC
2022 Class: Rivals=>Ohio is #12 in the MAC; 247Sports=>Ohio is #12
2023 Class: Rivals=>Ohio is #10 in the MAC; 247Sports=>Ohio is #4
2024 Class: Rivals=>Ohio is #5 in the MAC; 247Sports=>Ohio is #3

Here are links to various years on Rivals:
2021: https://n.rivals.com/team_rankings/2021/midam/football/re...
2022: https://n.rivals.com/team_rankings/2022/midam/football/re...
2023: https://n.rivals.com/team_rankings/2023/midam/football/re...
2024: https://n.rivals.com/team_rankings/2024/midam/football/re...
Rivals also gives comprehensive ratings for 2023, that include transfers (which moved Ohio down to #12):
https://n.rivals.com/team_rankings/2023/midam/football/co...

Just at a glance, it would seem that if Rivals ratings are historically correlated to results, Ohio should be in the bottom 3 in the MAC in wins, while 247Sports puts them more in the middle.

“We have two ears and one mouth so that we can listen twice as much as we speak.” ― Epictetus

Back to Top
  
L.C.
General User

Member Since: 8/31/2005
Location: United States
Post Count: 10,470

Status: Offline
  Message Not Read  RE: Using recruiting rankings to evaluate comparative talent levels
   Posted: 11/10/2023 8:41:42 PM 
FWIW, here are the Rivals ratings for Ohio 2024 recruits
5.7 Mathis
5.5 Erby, Jr., Kelly, Startz
5.4 Parson, Dorwart, Hurst, Anstead, Segarra, Thompson, Anderson
5.3 Brune, Bowman
NR Cemascoli, Ehret, Morton

In 2023, there were no players ranked over 5.5, three players rated 5.5, five players rated 5.4, five players rated 5.3, and four players got the bottom rating of 5.2.

“We have two ears and one mouth so that we can listen twice as much as we speak.” ― Epictetus

Back to Top
  
CoachPMac
General User

Member Since: 4/11/2023
Post Count: 32

Status: Offline
  Message Not Read  RE: Using recruiting rankings to evaluate comparative talent levels
   Posted: 11/13/2023 4:52:59 PM 
L.C. wrote:
CoachPMac wrote:
247 and On3 certainly have the ease of use features down. Rivals is still in the process of updating a system from the early 2000s. You can still view individual recruits with the recruit search feature, but it’s not as user friendly. Rivals will not create any more individual G5 team sites because they do not produce much money and in the end it’s all about the business.

The majority of evaluators from 5-10 years that made it seem like a “running joke” are now at either 247 or On3 btw.

I’ve added the 2023 and 2024 team rankings on the CSNbbs MAC main board for an easier viewing fashion.

Use whatever site suits you. I just wanted to touch on the Rivals part.

I did know that you could look up recruits individually at Rivals, though not at ESPN. I did not know that Rivals had restarted having MAC recruiting rankings, apparently back in 2021. I have no idea if Rivals or ESPN is now more accurate than they used to be at the G5 level. Both are, without question, accurate at the P5 level, and I'm not questioning that.

The main point of my post was specifically related to 247Sports. Since I have always considered them (along with Rivals, Scout.com, and ESPN) to be mostly random, I was surprised when I noticed that, not counting kickers, four True Freshmen are playing for Ohio, and all four are among the top 5 recruits in last year's class. In the past it was more normal for most of the True Freshmen who play to be at the bottom of the class. I therefore decided to do a test to see if overall MAC recruiting rankings at 247 were correlated to actual team wins, and to my surprise, they are.

Let's go back to that starting point, and see how Rivals and ESPN rated the four True Freshmen who are playing:
Chase Hendricks Rivals=3 stars/5.5, ESPN=3 stars/74, 247=86
Kobi Gorman Rivals=3 stars/5.5, ESPN=3 stars/76, 247=86
Ricky Hunt, Jr. Rivals=3 stars/5.5, ESPN=3 stars/72 247=86
Mason Williams Rivals=2 stars,5.4, ESPN=NR, 247=84

...and how they rated the other Ohio recruits from 2023:
Andrew Marshall Rivals=2 stars/5.4, ESPN=NR, 247=86
Danny Novickas Rivals=2 stars/5.4, ESPN=NR, 247=83
Jayvian Crable Rivals=2 stars/5.3, ESPN=NR, 247=82
Byron Pearson Rivals=2 stars/5.3, ESPN=NR, 247=82
Kendall Bannister Rivals=2 stars/5.2, ESPN=NR, 247=82
Jacob Winters Rivals=2 stars/5.4, ESPN=3 stars/72, 247=81
Desmond Duffy Rivals=2 stars/5.4, ESPN=3 stars/77, 247=80
Donovan Walker Rivals=2 stars/5.2, ESPN=NR, 247=80
Aidan Johnson Rivals=2 stars/5.3, ESPN=NR, 247=80
Owen DiFranco Rivals=2 stars/5.3, ESPN=NR, 247=80
Jalen Thompson Rivals=2 stars/5.3, ESPN=NR, 247=77
Brady Sestili Rivals=2 stars/5.2, ESPN=NR, 247=77
Gianni Spetic Rivals=2 stars/5.2, ESPN=NR, 247=75

Rivals looks very good. Their top three recruits (all 5.5) are all included in the 4 True Freshmen who are playing. The other true Freshmen who is playing is one of the 5 players that they rated next best, at 5.4. ESPN doesn't fare quite as well on this test. Their top-rated player, Desmond Duffy, is not playing, and one of the True Freshmen who is playing is NR. Still, of their five top rated players, 3 are playing, so there is still some correlation.

I did use ratings from the last five years to test the correlation between 247's rankings to team wins, and found a strong correlation. I would do the same for Rivals, but unfortunately I can not. MAC team rankings are only available back to 2021, and you need to go back further, because many/most starters are from the classed of 2019 or 2020, with even some from 2018 or before, due to Covid.

While 247Sports and Rivals do not completely agree, there are sometimes distinct differences, in this case, primarily related to the last two years. Here are team recruiting standings for Ohio in the MAC for the last four years. 2021 and 2022 they agree, but 2023 is quite different:
2021 Class: Rivals=>Ohio is #9 in the MAC; 247Sports=>Ohio is #9 in the MAC
2022 Class: Rivals=>Ohio is #12 in the MAC; 247Sports=>Ohio is #12
2023 Class: Rivals=>Ohio is #10 in the MAC; 247Sports=>Ohio is #4
2024 Class: Rivals=>Ohio is #5 in the MAC; 247Sports=>Ohio is #3

Here are links to various years on Rivals:
2021: https://n.rivals.com/team_rankings/2021/midam/football/re...
2022: https://n.rivals.com/team_rankings/2022/midam/football/re...
2023: https://n.rivals.com/team_rankings/2023/midam/football/re...
2024: https://n.rivals.com/team_rankings/2024/midam/football/re...
Rivals also gives comprehensive ratings for 2023, that include transfers (which moved Ohio down to #12):
https://n.rivals.com/team_rankings/2023/midam/football/co...

Just at a glance, it would seem that if Rivals ratings are historically correlated to results, Ohio should be in the bottom 3 in the MAC in wins, while 247Sports puts them more in the middle.


Yeah. It’ll take some time to prove how accurate the ratings are. But like I mentioned, Rivals is currently on top of MAC commitments/ratings and doing a better job than the other services in that area the past few years. Ease of use on the features still needs tuning.

What none of the sites do a good job of is taking into account players transferring out when it comes to wins. For example, I firmly believe Akron had the best overall class when it comes to portal/HS/JUCO this past recruiting class, but none of that matters when there is a lack of depth at QB and OL and the roster talent (outside of the previous class) was abysmal. Teams like Toledo, Ohio, Miami are so far ahead in layering and keeping talent. Kent’s classes the past few year’s wouldn’t match up to their win total this year because of all the talent they lost in the portal.

I don’t know how to post links here, but I did the 2023 MAC class including departures on the CSNbbs MAC main board and the 2024 MAC class is in progress there too.

Last Edited: 11/13/2023 4:53:44 PM by CoachPMac
Back to Top
  
Showing Replies:  1 - 8  of 8 Posts
Jump to Page:  1
View Other 'Ohio Football Recruiting' Topics
                                                                                                                                                                                                                                                                             



WHO'S ONLINE
There are 253 Active Visitors on BobcatAttack.com
253 Guests
0 Registered Users
There are no other users looking at this page.




Copyright ©2024 BobcatAttack.com. All rights reserved.  |  Privacy Policy  |  Terms of Use
Partner of USA TODAY Sports Digital Properties